Wireless Networked Optimal Control And Optimal Estimation Algorithms For Middle-size Soccer Robot Explicit Cooperation

In recent years with the rapid development of robot vision, intelligent control and information processing technology, the competition level of middle-size soccer robots in robot football World Cup (RoboCup) has been greatly improved. The efficient collaboration is required for the middle-size soccer robot game. It needs the perfect interaction among the robots to win the game. So a real-time and accurate communication environment is required in order to ensure the information exchange of the game between soccer robots. The wireless network is the designated communication method among soccer robots according to the rule of the game. However, the wireless network may bring random delays, distortion and even package loss during the information transmission process. To improve the cooperation efficiency between soccer robots, the interactive information should arrive at the receiving robots promptly and accurately. The main works are as follows in this thesis:Firstly, the middle-size soccer robot game rules are introduced, and the research necessity of the wireless networked control system in the soccer robot is discussed. Also, several wireless network protocols widely used in the soccer robot communication are presented, and in this thesis a reasonable wireless network for the soccer robot is selected from the aspect of the game’s practical requirement. Then main problems of the wireless networked control system in the soccer robots are discussed, and driving mode of the nodes in the wireless networked control system for the soccer robot has been chosen.Secondly, a specific scene is described in soccer robot explicit cooperation. Based on the scene, a real model of the wireless networked control system for the middle-size soccer robot is established with a consideration of the random time-delays, packet loss and outlier interference which existing in the measurement of system states. For outlier interference existing in the system state measurement, an anti-outlier stochastic optimal control algorithm, which adopts the improved kalman filter method combined with the stochastic optimal control, is proposed. Simulation results show that the algorithm can make the medium group soccer robots collaborate better and reduce the influence of the outliers effectively.Thirdly, considering that the proposed anti-outlier stochastic optimal control algorithm can only deal with time-varying delays and outliers, but the used zero-order holder can not realize better compensation for the packet loss, so a real-time queuing structure is adopted. Based on this structure, the optimal estimation compensator is designed for the packet loss. And the simulations show that this compensator has a good performance in solving the packet loss. Then the proposed algorithm in this thesis is compared with dynamic matrix control and predictive control algorithms respectively. The simulation results show that the proposed algorithm has a better control result.Finally, the results in this thesis are summarized, and the suggestion for further study is also given.